Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: Bill Jackson who wrote (73457)	3/5/2002 4:26:32 PM
From: combjelly	Read Replies (1) \| Respond to of 275872

"Yes, I expected that to be the case, but can you quantify that?" Not really. That is best left up to those who know more than I.

To: Bill Jackson who wrote (73457)	3/5/2002 5:01:32 PM
From: semiconeng	Read Replies (3) \| Respond to of 275872

Combjelly, Yes, I expected that to be the case, but can you quantify that? Bill Each extra layer adds an additional trip around the Fab. Since AMD seems to be using a Moto's Dual Damascene Cu Process, each additional layer would most likely require (NOT Necessarily in this specific order), the addition of an Insulating Dielectric, an Litho Pattern application for the Trench, an Etch step to cut the Trench, a Litho Pattern application for the Interconect Via's, an Etch step to cut the Vias in the dielectric, the insertion of the Via Plugs, the introduction of the copper metal, and a planerization. This does not include the additional Defect Metrology, and Critical Dimension Analytical steps. My WAG would be somewhere in the neighborhood of an additional 15-20 operations per layer, or approx 45-60 additional process steps for the 3 additional Metal Layers. Each additional step has the potential for additional defect introduction. Semi

To: Bill Jackson who wrote (73457)	3/6/2002 1:39:30 AM
From: Ali Chen	Read Replies (1) \| Respond to of 275872

Bill, "..but can you quantify that?" [9 layers of Hammer] According to classics, "Computer Architecture, a Quantative Approach" by J.Hennessy and D.Patterson, page 12-13, Die yield = Wafer yield x (1 + Defect per area x Die area /N)^-N , where N is "roughly the number of masking levels". In 1995, the good N was 3. Substituting "Defect per area" = 0.5 /cm2, Die = 1 cm2, and N=6, the yield in this model is 61.86%, while for N=9 it gives 61.47%. So, draw your own conclusion regarding impact of extra metal layers on Hammer yields. I wonder what is the function of BS vs. FUD amongst our friends, paid stooges? - Ali